Item — HN Reader

> For instance we teach arithmetic before calculus.

I don’t think that’s a fitting analogy.

Nearly everyone on the planet uses (basic) arithmetic. Very few use calculus.

By contrast, very few (programmers) use ASM, but nearly all of them use higher level languages.

1 comments

I'd say every programmer uses the constructs in assembly. Just because we have layers and layers of abstraction on top of that doesn't mean it's not valuable to understand the far simpler world that it all sits upon (Granted I understand it sits upon machine code, not assembly, but assembly is probably the closest to machine code that is more human interpret-able without significant effort).

My first language was BASIC on a V-tech. It's not quite the same but it still was such a fantastic starting point.

I've tried luring people into programming with Python for example and see them get frustrated by the amount of abstractions and indirection going on. I am really starting to like this idea of starting with assembly.

Yeah, my point wasn't that learning ASM isn't valuable, or that we don't use the constructs in higher level languages.

My point is that the analogy with arithmetic vs. calculus doesn't hold.

Nearly everyone uses basic arithmetic in everyday life, and a tiny fraction of those use calculus.

No programmer needs to learn ASM to be able to know how to use higher level languages. And a tiny fraction of them are using actual ASM in their everyday jobs.

Also, I think you can still learn the basic constructs of how languages work at a lower level without every learning actual ASM. There's no way you can learn calculus without an understanding of arithmetic.

Most people don't use basic arithmetic in everyday life anymore. They use machines which use arithmetic. Just like most programmers don't use assembly, they use programs which use assembly. In both cases, understanding what's going on is very useful even if you aren't directly touching that layer yourself.

By this token, everyone who counts apples in a market is using the axioms of Peano arithmetic every day.

The fact that our high level languages compile down to assembly doesn't mean we use assembly in any meaningful sense. My C code will be correct or not based on whether it conforms to the semantics of the C abstract machine, regardless of whether those semantics match the semantics of the assembly language that it happens to compile down to. Even worse, code that is perfectly valid in assembler may be invalid C, even if the C code compiles down to that same assembler code. The most clear example is adding 1 to an int variable that happens to have the value MAX_INT. This code will often compile down to "add, ax, 1" and set the variable to MIN_INT, but it is nevertheless invalid C code and the compiler will assume this value is impossible to happen.

This relationship between a programming language and assembler is even more tenuous for languages designed to run on heavy runtimes, like Java or JavaScript.